Unwanted fertility in Latin America: historical trends, … › ... ›...

193

Unwanted fertility in Latin America: historical trends, recent patterns1

John B. Casterline2

Jennifer Adams Mendoza3

Abstract This paper offers an overview of levels and trends in unwanted fertility in Latin America (including the Caribbean) based on national demographic surveys conducted from the mid-1970s to the present. We present estimates on a per child basis (percentage of births unwanted) and a per woman basis (unwanted births per woman, i.e. unwanted fertility rate). We apply to 59 WFS, DHS, and RHS surveys the estimation method recently proposed by Casterline and el-Zeini (2007); this method typically yields higher estimates of unwanted fertility than other available estimators. The contribution of this paper is to offer a portrait of unwanted fertility in the region that is revised, updated, and relatively comprehensive. There is considerable inter-country variability, but a summary of the general pattern is as follows. In the period since 2000, roughly one-third of births are unwanted, ranging from a low of 21% (Paraguay) to a high of 60% (Bolivia), and women experience on average about one unwanted birth during their reproductive career (synthetic cohort estimates), ranging from a low of 0.60 to a high of 2.5 unwanted births per woman. If these estimates are combined with Guttmacher estimates of induced abortion, the implication is that 1.5 – 2.0 unwanted pregnancies per woman is a common experience in the region (or, from a pregnancy perspective, about one-half of pregnancies are unwanted). The fertility decline experienced throughout the region in recent decades has consisted of different mixes of declines in wanted and unwanted fertility, no pattern predominates. Strikingly, wanted fertility in the recent period falls below replacement level (two births per woman) in almost all countries, and below 1.5 births per woman in a substantial minority of countries. However, the ideal number of children among young cohorts of women generally falls between 2.0 and 2.5, suggesting a contradiction between fertility ideals and fertility preferences (which take into account economic and social realities).

1 An earlier draft of this paper was presented at the annual meeting of the Population Association of America, Detroit, 30 April – 2 May 2008. The authors thank Suzana Cavenaghi for assistance with access to data, and Jose Miguel Guzman for comments on the earlier draft. Partial support for this research was provided by an award to the Initiative in Population Research at Ohio State from the National Institute of Child Health and Development (R21–HD47943). 2 Professor of Sociology, Ohio State University, Columbus, Ohio 43210 [email protected] PhD student in Sociology, Ohio State University, Columbus, Ohio, 43210. email: [email protected].

unwanteD fertility in latin america: historical trenDs, recent Patterns194

Introduction

There are multiple motivations for demographers’ continuing efforts to generate accurate estimates of the incidence of unwanted births (defined here as births not wanted at the time of conception). Prevention of unwanted births has long been a fundamental justification for investment of public and private resources in family planning services. Where reduction in population growth rates is a policy goal, the prevention of unwanted births can be a cost-effective step towards attainment of this goal, since unwanted births are assumed to be less costly to avert. The prevention of unwanted births also closes the gap between reproductive aspirations and outcomes, a worthy public policy goal in its own right. Finally, unwanted births are thought to carry distinctive and substantial costs (Gipson et al., 2008). These costs can consist of disadvantage suffered by the unwanted child – in health, in early childhood development, and in later social and economic opportunity – and of damage to the welfare of siblings, parents, and communities. The detrimental repercussions of unwanted childbearing are presumed to be multifaceted and potentially long-term.

This paper examines the current (and past) level of unwanted fertility in Latin America, as well as the contribution of changes in unwanted fertility to the substantial declines in fertility in the region that have occurred in the period from the 1970s to the present. The estimates of the incidence of unwanted births presented here have been calculated using a recently-developed method (Casterline, and el-Zeini, 2007) that we believe offers more valid estimates than the methods employed in past research on unwanted fertility in Latin America. We have also striven to maximize historical and cross-national coverage, analyzing surveys conducted as part of three major multi-national survey programs: World Fertility Survey (WFS), Demographic and Health Surveys (DHS), and Reproductive Health Surveys (RHS).

Demographers and public health researchers have long been aware of the relatively high levels of unintended pregnancy and unwanted birth in the Latin America and Caribbean region. An early multi-country study is Westoff’s (1981) six-country WFS analysis, in which Colombia and especially Peru stand out as having high unwanted fertility as compared to three Asian countries. In many subsequent multi-country studies, Peru, and to a lesser extent Colombia, have been identified as having distinctively high unwanted fertility (Blanc, 1982; Bongaarts, and Lightbourne, 1996; Westoff, and Moreno, 1996; Bongaarts, 1997; Adetunji, 1998). Analyses of DHS data have also pointed to high unwanted fertility in Bolivia (Westoff, and Moreno, 1996). Research that has taken a more global perspective has consistently concluded that the Latin American region experiences the highest rate of unwanted fertility (e.g.

John b. casterline anD Jennifer aDams menDoza 195

Bongaarts, 1997; Adetunji, 1998). Research in the 1980s and 1990s also concluded that unwanted fertility was on the rise in Latin America, and might well continue to rise, as a consequence of declines in the number of children desired (Westoff, 1981; Blanc, 1982; Bongaarts, 1997).

During the past decade, unwanted fertility has received less attention in comparative research on the region. Several overviews of fertility do not consider the distinction between wanted and unwanted fertility (Chakiel 2004, Guzman et al. 2006). A noteworthy exception is the comparative research of Ralph Hakkert (2001, 2003). Hakkert reviews the past research on unwanted fertility in the region and presents a thorough and revealing analysis of eight DHS surveys conducted in the mid- and late-1990s. Hakkert considers levels and trends in unwanted fertility and the correlates of unwanted fertility, while also being attentive to methodological issues. As such his piece is far more ambitious than our paper, although Hakkert’s research is now a bit dated.

Unwanted birth – the focus of this paper – is one of three possible outcomes of an unintended pregnancy, the other two being induced abortion and mistimed birth. Each of these outcomes is of concern in its own right, for reasons that are common to all three and specific to each one. All three can impact maternal and child health, with the impact possibly long-term (for the woman and, except for induced abortions, for the child). The social and economic consequences of unwanted births are in all likelihood substantially larger than the analogous consequences of the other two types of pregnancies. And unwanted births are distinctive in their effect on levels of fertility (and thereby population growth rates): unwanted births raise fertility, whereas mistimed births have minimal effect on fertility rates and induced abortions lower fertility. Therefore, while a comprehensive analysis of unintended pregnancy would encompass all three components, there are compelling reasons for examining each component separately. (While the empirical analysis in this paper is confined to unwanted births, our discussion in the final section incorporates regional estimates of induced abortion rates).

Our goal in this paper is to provide a portrait of unwanted fertility in the region that is revised, updated, and relatively comprehensive. We do not carry out empirical analysis of the determinants of unwanted fertility, but there is some value in listing the general categories of direct determinants as background to the descriptive portrait in this paper. A first point is that the risk of unwanted fertility is a function of desired fertility: everything else being equal, the fraction of women at risk of an unwanted birth is inversely related to the average number of children desired. Second, conditional on the fraction of women at risk, unwanted fertility is inversely related to the prevalence of contraceptive use among women who want


no more births (i.e. positively related to unmet need for stopping). This prevalence, in turn, is a function of the strength of the various obstacles to using contraception, including access to family planning services and a broad array of psychological, social, and cultural factors. Third, conditional on the prevalence of contraceptive use, unwanted fertility is inversely related to the effectiveness of contraception; that is, contraceptive failures can lead to an unwanted births. Fourth, conditional on the rate of occurrence of unwanted pregnancies, unwanted fertility is inversely related to the induced abortion rate. There are, then, four distinct direct determinants of population-level variation in the incidence of unwanted fertility, and any effort to explain cross-time or cross-country variation in unwanted fertility should account for all four determinants.

Method and Data

Method

Estimating the incidence of unwanted fertility is intrinsically a challenging task. At issue is the couples’ desires at the time of conception, but it is not practical to design data collection for national populations that ensures that interviews are conducted proximate in time to all (or even most) conceptions. Hence classification of births as wanted or unwanted will unavoidably depend on fertility desires measured prospectively or retrospectively, with the risk that the desires are not stable or, in the case of retrospective recall, incorrectly remembered. Adding to the challenge is the emotional sensitivity of the topic: respondents may feel that declaring a child “unwanted” is a violation of social or religious norms.

There are two widely-used methodologies for estimating the level of unwanted fertility, another method that has received little use, and the recently developed methodology that is applied in this research. The first method is retrospective direct inquiry, in which respondents are asked birth-by-birth about their preferences at the time of conception. The usual questionnaire item, asked of births during the previous three to five years, reads,

“At the time you became pregnant with <name>, did you want to become pregnant then, did you want to wait until later, or did not want (more) children at all?”

The chief virtue of this approach is that, in contrast to the other three methods, the variable of interest is directly measured, i.e. desires at the time of conception. However, it is now well-established that respondents are averse to reporting recent births (most of which will be living children) as “unwanted” (Bankole, and Westoff, 1998; Williams, and Abma, 2000). Hence estimates produced by this method are


known to be downwardly biased, and to a substantial extent in most settings. For

this reason, while the DHS collects this information, DHS reports do not use it for

estimates of wanted and unwanted fertility rates.

Instead the DHS uses the second method, popularized by Lightbourne (1985),

which relies on a comparison of the respondent’s ideal number of children and the

number of living children at the time of conception. The DHS wording of the key

item is,

“If you could go back to the time you did not have any children and could choose exactly the number of children to have in your whole life, how many would that be?”

If this ideal is equal to or less than the number of living children at the time of

conception of the birth in question, the birth is classified as unwanted. This method

produces higher estimates than the retrospective direct method (Bongaarts, 1990;

Casterline, and El-Zeini, 2007), suggestive of its greater validity. But the method has

serious shortcomings. For one thing, it relies on a survey item that is known to have

low test-test reliability (see studies cited in Casterline, and el-Zeini, 2007). Secondly,

two response patterns undermine the method: the tendency to report an ideal that

is no lower than the number of living children (so-called “rationalization”), and the

tendency in some societies to give a non-numeric answer (“up to God”). These two

response patterns both lead to downwardly-biased estimates of unwanted fertility.

Finally, there are valid reasons why the preference to have (or not have) more

children can be inconsistent with the ideal number of children (Bongaarts, 1990),

for example if the household is economically stressed or if sex preferences have not

been satisfied after the first few children.

Bongaarts (1990) proposes an aggregate estimator that relies on the prospective

preference item,

“Would you like to have (a/another) child, or would you prefer not to have any (more) children?”

This item has been shown to have higher test-retest reliability and higher validity

(by several criteria) than other standard fertility attitudinal items (see studies cited in

Casterline, and el-Zeini, 2007). Bongaarts’ method has, curiously, hardly been used in

the nearly twenty years since it was proposed. While it would appear to be superior

to the two methods described above, it also has significant disadvantages. Perhaps

the most important stem from the fact that this is a synthetic cohort approach: full

reproductive careers are inferred from cross-sectional patterns. That is, the method

assumes no inter-cohort differences in fertility desires (Hakkert, 2001). While

Bongaarts proposes a correction for this assumption, it is data-demanding and,

moreover, it is not clear that the correction is sufficient.


The final method is the “aggregate prospective” estimator recently introduced by Casterline and El-Zeini (2007). Like Bongaarts’ approach, this method relies on the prospective preference item which, as noted above, is known to have relatively high reliability and validity. Casterline and el-Zeini’s method is an aggregate method: it does not classify individual births as unwanted or unwanted, rather generates a birth-order-specific estimate of the proportion unwanted, with an estimate of the overall incidence of unwanted births calculated as a weighted average of the order-specific estimates. Unlike Bongaarts’ method, this is a strictly period estimator – order-specific prospective preferences at the time of the survey are used to estimate the fraction of births unwanted (by order) in a brief reference period preceding the survey. The data requirements are minimal, little more than prospective preferences at the survey and an accounting of births during the reference period. The basic formula is presented in equation (1):

up+1 = (1Np –

2vp) / Bp (1)

whereup+1 proportion of unwanted births in the reference period preceding the

survey

1Np proportion of women at the beginning of period who do not want

another child

2vp proportion of women at the survey who do not want another child and

did not have a child between the beginning of the reference period and survey

Bp proportion of women who had a birth in the reference period

And the overall proportion of births unwanted (u) is a weighted sum, as in equation (2):u = ∑ gp+1 up+1 (2)

wheregp+1 is the proportion of births in the reference period at order (p+1)

Age-specific estimates can be calculated on the basis of the order-specific estimates up+1 (i.e. age-by-age, weighted sums of the order-specific proportion unwanted are calculated), and, using these age-specific estimates, the unwanted total fertility rate (unwanted TFR) can be calculated. (See Casterline and el-Zeini (2007) for more details).1

1 A program in Stata for applying the Casterline – el-Zeini estimator to DHS and WFS data is available from the first author. Note that there are sample restrictions on each of the three elements of the formula given above, and these sample restrictions differ slightly among the three.


The “aggregate prospective” estimator assumes stability in fertility preferences during the reference period (in this application, 36 months). Preferences may be unstable in two directions, i.e. from “want more” to “want no more”, and vice versa. Of concern is instability among women who have a birth. If women state a preference to have no further births but change their mind before having a next birth, this will lead to upward bias in the estimate of unwanted fertility. If, in contrast, some women who state a preference to have another birth change their mind yet nevertheless have a birth, this will lead to downward bias. We have no empirical evidence on the relative weight of these two countervailing sources of bias, and indeed their relative weight probably varies by time and place. One might speculate that the first is more common, resulting in upwardly-biased estimates of unwanted fertility from the Casterline – el-Zeini aggregate prospective estimator. But in periods of rapid fertility decline, the second source of bias might well dominate, resulting in downwardly-biased estimates.

As expected, the aggregate prospective estimator produces higher estimates of unwanted fertility than the “Lightbourne method” (comparison of ideal and living children) that has been the primary method employed by demographers in recent years. (This is the method on which published DHS estimates are based.) Among the 59 surveys analyzed in this paper, on average the aggregate prospective estimate of the percentage of recent births unwanted is 16 percentage points higher (difference in medians) than the Lightbourne estimate, a substantial average difference.2

A final point concerns the interpretation of the wanted and unwanted TFRs that are presented later in this paper. As Hakkert (2001) observes, these TFRs are hybrid measures, a combination of preferences and fertility outcomes. They certainly are not pure representations of fertility desires. This is most clearly the case with the wanted TFR – some women do not have children they wish to have, for whatever reasons, and therefore fall short of their desired number. (We return to this point at the end of the analysis, when considering data on the ideal number of children.) Thus ordinarily desired fertility (usually unobserved) will exceed the wanted TFR. In contrast, the unwanted TFR actually represents the quantity of interest, provided there is confidence in the validity of the estimate. The reader is directed to Hakkert (2001) for a thorough and very insightful evaluation of the strengths and weaknesses of alternative estimators of wanted and unwanted fertility based on the standard battery of demographic survey items.

2 This comparison is complicated by the non-comparability across surveys in the measurement of the ideal number of children. Not all survey questionnaires included a fully appropriate item.


Data

Our ambition is to provide a comprehensive picture of levels and trends in

unwanted fertility in the Latin American region, including the Caribbean, for the

period from the onset of fertility decline to the present. To our knowledge we offer

the most expansive coverage across time and country of any research to date. Even

so, we analyze no surveys prior to the WFS surveys carried out in the late 1970s.

In future work we hope to enlarge our analysis by adding surveys from the 1960s.

But because the survey record from the 1960s is spotty, and because some Latin

American fertility declines started before the 1960s (i.e. the southern cone countries),

unfortunately the picture will remain incomplete.

We analyze survey data collected under three survey programs: the World Fertility

Survey (WFS), the Demographic and Health Surveys [DHS], and the Reproductive

Health Surveys (RHS). Two surveys were outside these three international programs:

the 2003 ENSAR in Mexico, and the 2006 PNDS in Brazil. We consider countries that

have at least one survey since 1990 -- fourteen countries and fifty-nine surveys in total.

Table 1 shows the surveys analyzed in this paper, by country, year, and survey program.

Table 1. Survey Data Analyzed in the Paper, by Country, Survey Program and Time Period

CountryDecade

1970s 1980s 1990s 2000s

South America Bolivia 1989 [D] 1993-94 [D] 1998 [D] 2003 [D] Brazil 1986 [D] 1996 [D] 2006 [P] Colombia 1976 [W] 1986 [D] 1990 [D] 1995 [D] 2000 [D] 2005 [D] Ecuador 1979 [W] 1987 [D] 1994 [R] 1999 [R] 2004 [R] Paraguay 1979 [W] 1990 [D] 1995-96 [R] 1998 [R] 2004 [R] Peru 1977 [W] 1986 [D] 1991-92 [D] 1996 [D] 2000 [D] 2004 [D]Central America Costa Rica 1976 [W] 1993 [R] El Salvador 1985 [W] 1993 [R] 1998 [R] 2002-03 [R] Guatemala 1987 [D] 1995 [D] 1998-99 [D] 2002 [R] Honduras 1996 [R] 2001 [R] 2005 [D] Mexico 1976 [W] 1987 [D] 2003 [E] Nicaragua 1992-93 [R] 1997-98 [D] 2001 [D] 2006-07 [R]Caribbean Dominican Republic 1975 [W] 1986 [D] 1991 [D] 1996 [D] 2002 [D] 2007 [D] Haiti 1977 [W] 1994-95 [D] 2000 [D] 2005-06 [D]

W = WFSD = DHSR = RHSP = PNDSE = ENSAR


A few comments on the country coverage:• The countries of the southern cone (Argentina, Chile, Uruguay) are entirely

missing from this analysis. The omission of Argentina is of particular concern, given its demographic weight.

• We include Costa Rica (surveys in 1976 and 1993) in some of the analysis, despite the absence of a recent survey.

• Several countries with WFS surveys (including Venezuela and Panama) are excluded because there has been no subsequent DHS or RHS survey since 1990.

• We are rather arbitrary in our inclusion of Caribbean countries, presenting estimates in this paper only for the Dominican Republic and Haiti. The justification for their inclusion is that both countries offer a time-series of surveys from the 1970s to the present.

As just noted, two Caribbean countries are included in this analysis. Rather than continually referring to “Latin America and the Caribbean” in the text, we use “Latin America” as a short-hand label.

The WFS and DHS surveys are relatively standardized across countries, although there are differences in the core questionnaires for the two survey programs that are relevant to research on unwanted fertility. Specifically: (1) The ideal number of children item in the WFS did not contain the prefatory clause “If you could go back to the time you did not have any children . . .”, a difference that is thought to result in higher ideal numbers in the WFS. (2) The WFS and DHS-I usually did not ask the retrospective child wantedness questions that became standard in later DHS phases.

In contrast to the WFS and DHS surveys, the questionnaires for the fifteen RHS surveys (conducted with technical assistance from the U.S. Centers for Disease Control) differ considerably from country to country, particularly in their fertility attitudinal items. For this analysis we have constructed extract files in which we attempted, within the constraints of the RHS questionnaires, to mimic the crucial DHS variables. In every survey we were able to construct a prospective preference variable that, in our judgment, is comparable to the WFS and DHS variables. This has not been feasible for ideal number of children nor for retrospective child wantedness. Fortunately our preferred estimator of unwanted fertility – the aggregate prospective estimator -- requires the prospective preference item. The greater variation in questionnaire design among RHS surveys threatens the validity of our comparative analysis. Offsetting this threat is the expanded coverage offered by the RHS, in particular Ecuador and Paraguay in South America and El Salvador, Guatemala, Honduras, and Nicaragua in Central America.

We generate estimates, using the aggregate prospective method, for births occurring in the thirty-six months preceding the survey (excluding births in the


month of interview). The estimates include births to never-in-union women but exclude births before age 15 (a very small fraction of total births in all countries).

Findings

All the estimates discussed in this section are listed in the Appendix.

Recent levels of unwanted fertility

We begin by examining the evidence on recent levels of unwanted fertility, as estimated from data in the most recent survey conducted since 2000 (thirteen countries -- Costa Rica is excluded).

In reviewing past literature in Section I, we noted that historically Latin America has experienced high unwanted fertility as compared to other regions. This remains the case. The median percentages of births unwanted for the most recent survey since 2000, by region, are in table 2.

Table 2

Region Median Percentage Unwanted Number Countries

Latin America & Caribbean 37 % 13

Sub-Saharan Africa 16 % 27

North Africa & West Asia 30 % 4

South & Southeast Asia 30 % 8

The country-specific fraction of births unwanted is shown in figure 1. This fraction varies widely within the region, from a high of 60 percent in Bolivia (2003 DHS) to a low of 21 percent in Paraguay (2004 RHS). A rather high incidence of unwanted births is the common pattern: in only two of the thirteen countries are less than one-quarter of births unwanted (Dominican Republic (24%) and Paraguay (20%)), whereas forty percent or more of births are unwanted in four countries (Bolivia (61%), Colombia (40%), Haiti (43%), and Peru (42%)). The estimates suggest that the incidence is especially high in the Andean countries (Bolivia, Peru), but also in the markedly different social and demographic setting of Haiti.

The exceptionally high percentage of births estimated for Bolivia stands out in figure 1. We have investigated this further. A first point is that a high incidence of unwanted fertility especially characterizes the indigenous sub-groups, who contribute about two-thirds of the Bolivian births: the incidence for all indigenous groups combined is 66 percent, as compared to 51 percent for the non-indigenous portion of the Bolivian sample. To be sure, the latter is a high rate as compared


to national figures for other countries, but clearly it is the incidence among the indigenous groups in Bolivia that is exceptional.3 It is outside the bounds of this research to ascertain the reasons for this very high level of unwanted fertility, but we can comment briefly on what the existing research literature suggests. Indigenous women are less likely to use any method of contraception (Mendoza, and Heaton, 2008). This differential undoubtedly is due in part to lower socioeconomic status, but it may also reflect barriers to accessing services that are specific to these groups, such as greater geographic distance, language barriers, and distrust of non-indigenous medicine (Schuler et al., 1994; Terborgh et al., 1995). And among those women who do practice contraception, indigenous women are more likely to use less effective methods (Mendoza, and Heaton, 2008), increasing the likelihood of contraceptive failure. A final factor that could account for the much higher incidence of unwanted fertility among indigenous women in Bolivia would be a lower likelihood of resorting to induced abortion once an unwanted pregnancy occurs. We are not aware of any empirical literature that examines this.

Figure 1. Percentage of Births Unwanted Most Recent Survey Since 2000

Note: Aggregate Prospective estimates, births during 36 months before survey.

3 The estimate for the indigenous portion of the Peruvian sample is identical – 66%. This provides some reassurance about the validity of the Bolivian estimates. Note that the indigenous sub-groups carry far less demographic weight in Peru, contributing only 15% of births.


Estimates of the unwanted Total Fertility Rate are shown in figure 2. These can be interpreted as the average number of unwanted births women would have over the course of their reproductive careers if the age-specific rates of unwanted fertility observed in the thirty-six month reference period prevailed throughout their reproductive years. Most striking are the very high rates in Bolivia and Haiti, at least two births per woman (note also the rate in Guatemala of more than 1.5 births per woman). Although these are exceptional cases, the rate equals or exceeds one birth per woman in eight of the thirteen countries; those under one birth per woman are Brazil (0.6), Dominican Republic (0.6), El Salvador (0.9), Nicaragua (0.8), and Paraguay (0.7). Levels of contraceptive sterilization are relatively high in these countries, except for Paraguay, and undoubtedly this is an important proximate cause of their relatively low unwanted fertility rates. The unwanted TFR for nine of the thirteen countries falls between 0.7 and 1.4; thus one could say, speaking in most general terms, that the common regional experience in this recent period has been roughly one unwanted birth per woman.

Figure 2. Unwanted Total Fertility Rate Most Recent Survey Since 2000

Note: Aggregate Prospective estimates, births during 36 months before survey.

It follows that unwanted fertility contributes substantially to the overall level of fertility in the region. This is confirmed in figure 3, which shows the TFR as the sum of its two components (wanted and unwanted TFR). On average (median) the unwanted TFR constitutes 43 percent of the TFR in these thirteen countries. Only in the Dominican Republic and Paraguay does the unwanted TFR make up less than 30 percent of the TFR, and in Bolivia and Haiti one-half or more of the TFR consists of unwanted fertility.


Figure 3. Total Fertility Rate and Components Most Recent Survey Since 2000

Note: Aggregate Prospective estimates, 36 months before survey.

Trends in unwanted fertility and the Latin American fertility decline

Trends in both the percentage of births unwanted and the unwanted TFR are shown in figures 4a–4c. These figures are a bit confusing because of the different scales for the percentage of births unwanted (solid line and left-hand scale) and the unwanted TFR (dashed line and right-hand scale). But this layout has the advantage of facilitating comparison of the pair of trends in each country. Note that there is no necessary association between the two trends; for example, when fertility declines the unwanted TFR may also decline despite stability or even increase in the percentage of births unwanted.

A variety of patterns of reproductive change are evident in figures 4a–4c. The most common pattern is decline in both the percentage of births unwanted and the unwanted TFR. This is the pattern in the six South America countries. It is also the pattern in three of the six Central American countries – El Salvador, Mexico, and Nicaragua, although in El Salvador the decline is slight (Figure 4b). And this pattern occurs to a dramatic extent in the Dominican Republic, and there is some indication that this is the emerging pattern in Haiti (Figure 4c).

But the preceding summary gives an exaggerated impression of the extent to which the two variables march down together. Another salient feature of figures 4a–4c is the slower pace of decline in the percent of births unwanted. This observation applies to Bolivia, Colombia, El Salvador, Mexico, Dominican Republic, and Haiti.


In these two South American countries (Bolivia and Colombia), the percentage of

births unwanted has remained relatively stable while the unwanted fertility rate has

fallen substantially (by at least one birth per woman).

Figure 4a. Trends in Unwanted Fertility: South America


Figure 4b. Trends in Unwanted Fertility: Central America



Figure 4c. Trends in Unwanted Fertility: Caribbean


From the standpoint of women and households, it is probably of more importance that the rate of unwanted childbearing (i.e. births per woman) declines: everything else being equal, as the volume of unwanted births declines, the social and economic costs should ease. And these individual- and household-level gains probably aggregate up to societal gains, not to mention further macro returns from lower fertility and slower population growth. But from the standpoint of children, the fact that the fraction of births unwanted hardly declines is of some concern. That is, from a birth cohort perspective, the prevalence of unwantedness (at conception) has been relatively fixed in these countries over the past few decades (with, to be sure, some notable exceptions, namely Dominican Republic, Nicaragua, and Peru). Furthermore, the societal context has been far from stable, rather has been quite dynamic, including economic changes that have placed a higher premium on educational credentials and on women working away from the home. These changes may have increased the costs of unwanted childbearing. Hence it is altogether possible that, despite the lower rates per woman of unwanted fertility in the recent period, the individual- and societal-level impact of unwanted fertility matches or even exceeds the impact of the higher rates in the past.

A final question about trends concerns the relative contribution of declines in wanted and unwanted fertility to the overall fertility decline. This question is addressed in figures 5a-5c and table 3. Evidently these Latin American declines


have been produced by different mixes of declines in the two components. The decomposition in table 3 makes this point most clearly: The fraction of the overall fertility decline due to declines in unwanted fertility has ranged from a high of 90 percent in Nicaragua (where wanted fertility has remained rather stable at two births per woman from the early 1990s to the present) to a low of 18 percent in Haiti (where unwanted fertility remained over two births per woman from the late-1970s until the most recent survey). The relative contributions of declines in wanted and unwanted fertility vary so much as to resist any generalization. In attempting to perceive common patterns across countries in figures 5a-5c and table 3, it should be kept in mind that the time-period encompassed differs considerably from country-to-country: in some countries the period is no more than two decades, whereas in other countries only the most recent 15 years or so is examined. Were the equivalent analysis to be performed in all countries on the entire period of fertility decline, it is possible that a discrete number of common patterns might emerge.

Figure 5a. Trends in Fertility: South America



Figure 5b. Trends in Fertility: Central America


Figure 5c. Trends in Fertility: Caribbean



Table 3. Fertility Decline: Contributions of Declines in Wanted and Unwanted Fertility

Country Survey YearsAmount Decline

in TFRPercentage Contribution

Wanted Unwanted TotalSouth America Bolivia 1989, 2003 -1.20 9.3 90.7 100 Brazil 1986, 2006 -1.64 40.7 59.3 100 Colombia 1976, 2005 -2.14 29.8 70.2 100 Ecuador 1979, 2004 -2.00 34.7 65.3 100 Paraguay 1979, 2004 -1.94 69.9 30.1 100 Peru 1977, 2004 -2.77 28.1 71.9 100Central America Costa Rica 1976, 1993 -0.48 55.8 44.2 100 El Salvador 1985, 2003 -1.41 55.1 44.9 100 Guatemala 1987, 2002 -1.29 71.5 28.5 100 Honduras 1996, 2005 -1.97 61.4 38.6 100 Mexico 1976, 2003 -3.23 35.1 64.9 100 Nicaragua 1992, 2006 -1.90 9.7 90.3 100Caribbean Dominican Republic 1975, 2007 -3.01 24.3 75.7 100 Haiti 1977, 2005 -1.57 82.2 17.8 100

Current levels of wanted fertility and the future course of Latin American fertility

While the focus of this research is unwanted fertility, estimates of wanted fertility are a by-product, and these are of interest in their own right. The wanted TFRs for the most recent survey since 2000 are shown in figure 6. These are the total number of births women would have, on average, if unwanted fertility were entirely eliminated. To reiterate, this is a synthetic cohort estimate based on fertility rates in the three years preceding the survey. It should not be viewed as a measure of desired fertility, because some women may have failed to have births that were desired.

In eleven of the thirteen countries – Guatemala and Paraguay are the exceptions – the wanted TFR is less than replacement-level, i.e. less than two births per woman. Indeed, in four countries the wanted TFR is less than 1.5 births per woman: Brazil, Bolivia, Colombia, and Peru. Moreover, it is evident from figure 5a that, according to our estimates, rather low levels of wanted fertility have characterized these South American countries for several decades, in the cases of Colombia and Peru as far back as the mid-1970s. Perhaps most surprising are the low levels of wanted fertility in Bolivia and Peru in the past when their overall TFRs exceeded four births per woman and when social and economic conditions were relatively undeveloped. There appears to be an anti-natalism in these Andean societies that is rather deeply rooted.4

4 A separate analysis of the Peruvian surveys limited to the Sierra region -- which resembles Bolivia in its social, cultural, and economic features – also reveals consistently low wanted TFRs since the 1980s.


Figure 6. Wanted Total Fertility Rate Most Recent Survey Since 2000


What are the implications for the future of the estimates in figure 6? As evident in figure 3, at present the overall TFR – the sum of wanted and unwanted fertility -- exceeds two births per woman in all countries included in this analysis. A future with no unwanted fertility is implausible -- in no society has unwanted fertility been entirely absent. And, indeed, from the record of the past three decades one could infer that Latin American societies are especially susceptible to unwanted fertility, for whatever mix of reasons. Having said this, if one anticipates more perfect birth control in the future and consequent reductions in unwanted fertility – certainly a desirable outcome on social and public health grounds – then from the estimates in figure 6 one could reasonably posit that post-transition rates of fertility in the region will generally lie below replacement level (i.e. less than two births per woman).

A different perspective on likely future levels of fertility is provided by survey data on women's ideal number of children. The mean ideal for young cohorts (women age 15-24) is presented in figure 7a, and the percentage expressing an ideal of 0 or 1 child is presented in figure 7b. These data indicate little attitudinal support for sub-replacement fertility. The mean ideal exceeds 2.0 children in all countries except Brazil and Colombia, where the mean is 1.9 children. In five countries (Dominican Republic, Guatemala, Haiti, Honduras, Paraguay) the mean ideal equals or exceeds 2.5 children. And it is evident in figure 7b that relatively small fractions of women express an ideal of 0 or 1 child – only in Brazil, Bolivia and Colombia does this amount to one-fifth or more of these younger women.


Figure 7a. Ideal Number of Children, Women Aged 15-24 Most Recent Survey Since 2000

Note: Among women who provide a numeric ideal.

Figure 7b. Ideal Number of Children Equals 0 or 1, Women Aged 15-24 Most Recent Survey Since 2000

Note: Among women who provide a numeric ideal.

As guides to future levels of fertility, the recent unwanted TFR and the ideals of younger women are in contradiction, with the latter suggesting substantially higher fertility (one-half child or more) than the former. This is a common predicament in


contemporary low fertility societies, i.e. achieved period fertility falling well short of stated ideals, especially the ideals expressed by younger cohorts (Morgan, and Taylor, 2006). The conventional interpretation of this apparent contradiction is that achieved fertility falls short of ideals because couples defer to various economic and social constraints. The estimates in figures 6 and 7 indicate that the same predicament is coming to characterize Latin American societies. Indeed, because the wanted TFR must be no greater than the TFR and typically will be lower than the TFR (because of the existence of unwanted fertility), it follows that the wanted TFR will fall short of the mean ideal number of children so long as the ideal at least equals the TFR, which is often the case in low and moderate fertility societies. We note again that the wanted TFR is not a pure measure of fertility desires, rather mixes desires and actual outcomes (with the latter possibly affected by voluntary and involuntary factors that depress fertility).

Summary and Concluding Comments

This is a study of reproductive change that takes advantage of the large number of national demographic surveys conducted during the past four decades in Latin America and the Caribbean. We give special attention to the most recent decade, but also examine trends since the mid-1970s. The historical analysis provides the context for consideration of recent estimates, and also yields some insights about the nature of fertility decline in the region.

Our focus in this paper has been unwanted fertility, viewed both from a child perspective (percentage of births unwanted) and from a woman perspective (rate per woman). Those familiar with the demography of the region will not be surprised that we estimate high rates of unwanted fertility in the region even in the recent period – high according to absolute standards (roughly one unwanted birth per woman on average, and well in excess of this rate in some countries) and high as compared to the incidence of unwanted births in other major regions. This is a familiar outcome -- survey data as far back as the 1970s indicated that rates of unwanted fertility were relatively high in Latin America, and evidently the region has maintained this standing up to the present.

This is not to deny that downward trends in unwanted fertility are not also apparent. While the fraction of births unwanted remains rather stable in most of the countries examined (typically around 30%-40% of births), the unwanted fertility rate – the number of unwanted births per woman -- has fallen substantially in most countries. Whether the individual and societal costs of unwanted childbearing have declined by an equivalent amount cannot be assumed. One could reasonably surmise that per child costs of unwanted fertility are higher now than in the past.


As noted at the outset, unwanted birth is one of three components of unintended pregnancy, the other two being induced abortion and mistimed birth. Induced abortion rates are notoriously difficult to estimate accurately. Perhaps the most trustworthy and cross-nationally comparable estimates have been generated by researchers at the Guttmacher Institute (Sedgh et al. 2007). Their most recent estimates are centered on 2003, and are shown in table 4.

Table 4

Abortion Rate

(per 1000 women)

Abortion Ratio

(per 100 births)

South America 33 38

Central America 25 26

Caribbean 35 42

Multiplying the abortion rates by twenty-five (an approximation of the number of years in an average reproductive career) yields a lifetime average number of abortions ranging from 0.6 per woman (Central America) to 0.9 per woman (Caribbean). If these are added to the unwanted TFRs shown in figure 2, most of which approach or exceed 1.0 per woman, and if one assumes lifetime exposure to recent rates, the conclusion is that the average regional experience is 1.5 – 2.0 unwanted pregnancies per woman, a remarkably high rate. Alternatively, considering this on a per child basis, if one were to take the regional median of 37 unwanted births per 100 births (see Appendix and table 2 earlier in this paper) and use Guttmacher’s South American abortion ratio of 38 abortions per 100 births, the two combined imply 54 unwanted pregnancies per 100 pregnancies.5 That is, the common experience in the region would appear to be one-half or more of pregnancies unwanted (and subsequently leading either to an induced abortion or an unwanted birth). This is stark evidence of a large discrepancy between reproductive aspirations and achievements.

Our aim in this paper has been to present a revised and up-to-date set of estimates of unwanted fertility for the maximum number of countries in the Latin American region. As such this is a piece of descriptive analysis, for which we do not apologize: good description of demographic realities, especially demographic realities that have major public policy implications, is among the chief responsibilities of the demographic research community. But certainly this research should be followed by investigation of the determinants of unwanted fertility, as well as assessments of policy options for reducing it. If success in reducing unwanted fertility were to generate a new set of concerns that fertility is too low, then this can become the focus of a different set of research endeavors and public policy evaluations. There are

5 Calculated as (37 + 38) / (100 + 38).


costs attached to unwanted fertility (for the child, for the parents). If sub-replacement fertility is viewed as a societal problem, unwanted fertility is a sub-optimal solution.

References

Adetunji, Jacob A. (1998). “Unintended childbearing in developing countries: levels, trends, and determinants.” DHS Analytical Reports No. 8. Calverton, MD: Macro International.

Bankole, Akinrinola and Charles F. Westoff (1998). “The consistency and validity of reproductive attitudes: evidence from Morocco,” Journal of Biosocial Science 30(4): 439-455.

Blanc, Ann Klimas (1982). “Unwanted fertility in Latin America and the Caribbean.” International Family Planning Perspectives 8(4): 156-162.

Bongaarts, John (1990). “The measurement of wanted fertility.” Population and Development Review 16(3): 487-506.

_________ (1997). “Trends in unwanted childbearing in the developing world.” Studies in Family Planning 28(4): 267-277.

_________ and Robert E. Lightbourne (1996). “Wanted fertility in Latin America: trends and differentials in seven countries.” In J.M. Guzman et al. (eds) The Fertility Transition in Latin America. New York: Oxford University Press.

Casterline, John B. and Laila O. el-Zeini (2007). “Estimation of unwanted fertility.” Demography 44(4): 729-745.

Chackiel, Juan (2004). “La transición de la fecundidad en América Latina 1950-2000.” Papeles de Población 41: 9-58.

Gipson, Jessica D., Michael A. Koenig, and Michelle Hindin (2008). “The effects of unintended pregnancy on health outcomes: a review of the literature.” Studies in Family Planning 39(1): 18-38.

Guzmán, José Miguel, Jorge Rodríguez, Jorge Martínez, Juan Manuel Contreras and Daniela González (2006). “The demography of Latin America and the Caribbean since 1950.” Populations 61( 5-6): 519-620.

Hakkert, Ralph (2001). “Levels and determinants of wanted and unwanted fertility in Latin America.” Paper presented at the General Conference of the IUSSP, Salvador, Brazil, 18-24 August.

_________ (2003). “Fecundidad deseada y no deseada en América Latina, con particular referencia a algunos aspectos de género.” Pp.267-288 in La fecundidad en América Latina: ¿Transición o revolución? Santiago, Chile: CEPAL.

Lightbourne, Robert E. (1985). “Individual preferences and fertility behaviour.” Pp. 165-198 in John Cleland and John Hobcraft (eds) Reproductive Change in Developing Countries: Insights from the World Fertility Survey. Oxford: Oxford University Press.

Mendoza, Jennifer Adams and Tim B. Heaton (2008). “The incomplete contraceptive revolution in Bolivia.” Paper presented at the annual meeting of the American Sociological Association, Boston, August 1-4 2008.


Morgan, S. Philip and Miles G. Taylor (2006). “Low fertility at the turn of the twenty-first century.” Annual Review of Sociology 32: 375-399.

Sedgh, Gilda, Stanley Henshaw, Susheela Singh, Elisabeth Ahman and Iqbal H. Shah (2007). “Induced abortion: estimated rates and trends worldwide.” The Lancet 370: 1338-1345.

Shuler, Sidney Ruth, Maria Eugenia Choque and Susanna Rance (1994). “Misinformation, Mistrust, and Mistreatment: Family Planning among Bolivian Market Women.” Studies in Family Planning 25(4): 211-221.

Terborgh, Anne and James E. Rosen, Roberto Santiso Galvez, Willy Terceros, Jane T. Bertrand, and Sheana E. Bull (1995). “Family planning among indigenous populations in Latin America.” International Family Planning Perspectives 21(4): 143-149, 166.

Westoff, Charles F. and Lorenzo Moreno (1996). “Reproductive intentions and fertility in Latin America.” In J.M. Guzman et al. (eds) The Fertility Transition in Latin America. New York: Oxford University Press.

_________ (1981). “Unwanted fertility in six developing countries.” International Family Planning Perspectives 7(2): 43-52.

Williams, Lindy and Joyce Abma (2000). “Birth wantedness reports: a look forward and a look back.” Social Biology 47(3-4): 147-163.


Appendix. Unwanted Fertility Estimates

Country Year Survey Program

Total Fertility RatesPercent of Births

Unwanted Overall TFR

Unwanted Wanted

Ideal vs. Living

Aggregate Prospective Ideal Bongaarts Aggregate

ProspectiveAggregate

ProspectiveBolivia 1989 DHS 41.4 67.4 5.04 2.28 3.26 3.56 1.47

Bolivia 1993 DHS 38.3 59.2 4.77 2.02 2.89 3.02 1.74

Bolivia 1998 DHS 37.6 59.5 4.23 1.75 2.49 2.70 1.53

Bolivia 2003 DHS 41 60.2 3.84 1.74 2.26 2.48 1.36

Brazil 1986 DHS 29.2 43.1 3.42 1.15 1.37 1.60 1.82

Brazil 1996 DHS 28.7 44.4 2.54 0.75 1.04 1.16 1.37

Brazil 2006 PNDS 21.6 34.1 1.79 0.39 0.43 0.63 1.16

Colombia 1976 WFS 21.2 47.3 4.54 1.21 1.83 2.51 2.02

Colombia 1986 DHS 27.9 48.4 3.20 1.06 1.27 1.73 1.47

Colombia 1990 DHS 23.2 37.4 2.82 0.75 0.77 1.18 1.64

Colombia 1995 DHS 25 39.9 2.97 0.81 1.20 1.27 1.71

Colombia 2000 DHS 29 39.7 2.61 0.81 1.04 1.10 1.51

Colombia 2005 DHS 28.7 40.2 2.39 0.73 0.91 1.01 1.38

Costa Rica 1976 WFS 15.6 31.8 3.58 0.63 0.72 1.29 2.29

Costa Rica 1993 RHS 15.2 33.1 3.10 0.52 0.77 1.08 2.02

Dominican Republic

1975 WFS 21.3 44.2 5.45 1.53 2.37 2.90 2.55

Dominican Republic

1986 DHS 23 39.1 3.68 1.09 1.19 1.71 1.98

Dominican Republic

1991 DHS 19.6 31.2 3.34 0.76 0.87 1.19 2.15

Dominican Republic

1996 DHS 19.6 25.9 3.17 0.70 0.94 0.90 2.26

Dominican Republic

2002 DHS 19.8 29 2.99 0.63 0.88 0.93 2.06

Dominican Republic

2007 DHS 19.6 23.6 2.43 0.51 0.45 0.62 1.82

Ecuador 1979 WFS 17.5 41.7 5.18 1.18 1.99 2.56 2.62

Ecuador 1987 DHS 29.3 52.1 4.24 1.42 2.05 2.41 1.83

Ecuador 1994 RHS 24.8 40.8 3.49 1.02 1.46 1.66 1.83

Ecuador 1999 RHS 28.9 44 3.30 1.07 1.41 1.61 1.69

Ecuador 2004 RHS 26.5 35.9 3.18 0.92 1.09 1.25 1.93

El Salvador 1985 DHS 12.8 34.6 4.21 0.70 1.31 1.57 2.64

El Salvador 1993 RHS 1.2 30.4 3.79 0.06 1.14 1.37 2.42

El Salvador 1998 RHS 1.2 32.5 3.51 0.05 1.14 1.31 2.20

El Salvador 2003 RHS 0.7 29.4 2.81 0.02 0.77 0.94 1.87

Guatemala 1987 DHS 16.6 32.4 5.59 1.09 1.87 1.98 3.61

Guatemala 1995 DHS 18.4 35.1 5.13 1.10 1.84 1.97 3.16

Guatemala 1998 DHS 15.9 37.7 5.04 0.94 2.11 2.09 2.95

Guatemala 2002 RHS 16 33.4 4.31 0.80 1.37 1.61 2.69

(continued)


Country Year Survey Program

Total Fertility RatesPercent of Births

Unwanted Overall TFR

Unwanted Wanted

Ideal vs. Living

Aggregate Prospective Ideal Bongaarts Aggregate

ProspectiveAggregate

ProspectiveHaiti 1977 WFS 20.1 33.9 5.48 1.44 1.83 2.26 3.22

Haiti 1994 DHS 30.5 44.2 4.78 1.75 2.18 2.45 2.33

Haiti 2000 DHS 33.7 45.6 4.69 1.90 2.38 2.48 2.21

Haiti 2005 DHS 31.8 42.5 3.92 1.53 1.66 1.98 1.93

Honduras 1996 RHS 26.9 34.5 5.23 1.74 2.25 2.16 3.07

Honduras 2001 RHS 24.2 30 4.39 1.26 1.61 1.55 2.85

Honduras 2005 DHS 24.7 37.9 3.26 0.93 1.13 1.40 1.86

Mexico 1976 WFS 22.3 46.2 5.86 1.61 2.80 3.12 2.74

Mexico 1987 DHS 25.6 47.2 4.02 1.17 1.64 2.08 1.94

Mexico 2003 ENSAR 15 37.3 2.63 0.42 0.64 1.02 1.60

Nicaragua 1992 RHS 74.3 55.6 4.55 3.49 2.39 2.56 1.98

Nicaragua 1997 DHS 27.6 50.6 3.63 1.17 1.75 2.01 1.62

Nicaragua 2001 DHS 25.2 42.1 3.23 0.96 1.33 1.52 1.70

Nicaragua 2006 RHS 15.4 27.4 2.65 0.48 0.54 0.85 1.80

Paraguay 1979 WFS 7.5 21.7 4.87 0.49 1.04 1.33 3.55

Paraguay 1990 DHS 13.4 30.4 4.71 0.73 1.29 1.63 3.08

Paraguay 1995 RHS 17.5 31.8 4.07 0.79 1.14 1.44 2.63

Paraguay 1998 RHS 9.4 34.4 4.26 0.45 1.26 1.60 2.66

Paraguay 2004 RHS 14.6 20.5 2.93 0.52 0.60 0.74 2.19

Peru 1977 WFS 28.1 51.3 5.28 1.79 3.11 2.17

Peru 1986 DHS 38.1 66.2 4.12 1.79 2.30 2.89 1.22

Peru 1991 DHS 38.4 55.5 3.54 1.54 1.77 2.16 1.38

Peru 1996 DHS 34.2 53.6 3.54 1.35 1.87 2.05 1.48

Peru 2000 DHS 32.9 46.8 2.85 1.03 1.19 1.44 1.41

Peru 2004 DHS 31.8 41.7 2.51 0.85 0.86 1.11 1.39

(continued)

Date post:	05-Jul-2020
Category:	Documents
Upload:	others
View:	0 times
Download:	0 times

Unwanted fertility in Latin America: historical trends, … › ... ›...

Documents